Training Platforms

All Novatech Deep Learning systems have been designed with performance and stability in mind. All our systems go through detailed Q+A as well as Q+C inspections before being deployed, typically this is done over 24-48 hours. Training systems are defined by the users need for far higher levels of GPU compute, Scalability, and accessibility. Typically, the experiments users are running are of a more complicated nature and using much larger data sets. These systems are normally always rack mounted and would normally use a minimum of x4 GPU’s.

  • Novatech DL-TR1

    • GPUs

      2x Tesla V100

    • CPU

      2x Intel Scalable Gold 5118 2.3Ghz

    • Storage

      Data:2x 2TB
      OS: 1x 480GB SSD

    • Memory

      192GB 2666Mhz

    • Network

      2x Intel 1Gb LAN

    View platform
  • Novatech DL-TR2

    • GPUs

      4x NVIDIA TESLA V100

    • CPU

      2x Intel Gold 5118 12 Core 2.3Ghz

    • Storage

      Data: 3x 1.92 TB SSD RAID 0
      OS: 1X 1.92 TB SSD

    • Memory

      256 GB LRDIMM DDR4

    • Network

      Dual 10GBASE-T (RJ45)

    View platform
  • Novatech DL-TR3

    • GPUs

      2x NVIDIA TESLA V100

    • CPU

      2x Intel E5-2628L V4 1.9Ghz

    • Storage

      Data:2x 2TB
      OS: 1x 480GB SSD

    • Memory

      256GB 2400Mhz

    • Network

      2x Intel 10Gb LAN ports

    View platform
  • Novatech DL-TR4

    • GPUs

      8x NVIDIA TESLA V100

    • CPU

      2x Intel Xeon Gold 5120 2.2Ghz

    • Storage

      Data:2x 2TB
      OS: 1x 480GB SSD

    • Memory

      128GB 2666Mhz

    • Network

      2 x 10Gb/s BASE-T LAN portss

    View platform
  • Novatech DL-TR5

    • GPUs

      4x NVIDIA TESLA V100 32GB

    • CPU

      1x AMD Epyc 7551P - 32 CoreS | 64 Threads

    • Storage

      Data: 2x 2 TB SSD
      OS: 1X 480GB SSD

    • Memory

      128GB RDIMM DDR4

    • Network

      Dual 10GBASE-T (RJ45)

    View platform

Latest articles